|Publication number||US4509015 A|
|Application number||US 06/418,641|
|Publication date||Apr 2, 1985|
|Filing date||Sep 16, 1982|
|Priority date||Sep 21, 1981|
|Also published as||DE3271562D1, EP0076054A1, EP0076054B1|
|Publication number||06418641, 418641, US 4509015 A, US 4509015A, US-A-4509015, US4509015 A, US4509015A|
|Inventors||Roger J. Ordidge, Peter Mansfield|
|Original Assignee||Ordidge Roger J, Peter Mansfield|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (2), Referenced by (55), Classifications (9), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This invention relates to methods and apparatus for investigating a body by nuclear magnetic resonance (NMR). Such methods and apparatus have application in obtaining images of non-homogeneous bodies and much interest has been aroused in applying these techniques to obtaining images of biological tissue and even living subjects. The technique provides an alternative facility to the use of X-rays and is not prone to the known dangers of exposure of living matter to X-rays.
NMR methods involve the excitation of nuclear spins within a body by the use of magnetic fields and radio frequency pulses. Signals from the free induction decay of spins are read out and may be subject to Fourier transformation in order to obtain image information. To obtain spatial discrimination a main magnetic field which is static may have superimposed thereon a magnetic field gradient. An example of such a method using switched magnetic field gradients is described in British patent specification No. 1,596,160. In that specification there is described a method of obtaining signals representing the nuclear magnetic resonance spin density distribution in a body which comprises preferentially exciting nuclear magnetic resonance in a slice of a body, applying a magnetic field to said body with gradients in said slice in different directions and repetitively reversing the sense of one of said gradients at a rate which allows cyclic regrowth and decay of the free induction decay (FID) signal from the sample, and reading out the resulting FID signal.
It is an object of the present invention to provide an improved form of such a method of investigating a body by nuclear magnetic resonance.
According to one aspect of the present invention, a method of investigating a body by nuclear magnetic resonance comprises: preferentially exciting nuclear magnetic resonance in a slice of the body; applying for a first period a magnetic field to said body with first and second gradients in first and second directions respectively in said slice, the sense of said first gradient being periodically reversed; re-exciting said slice and re-applying said magnetic field for a second period but with the first gradient periodically reversed in sense with opposite phase to that obtaining during said first period; detecting the resultant free induction decay signals occurring during each application of said magnetic field; editing the detected signals to obtain a set of signals occurring when said first gradient has one sense; and processing said set of signals to retrieve data.
According to a second aspect of the present invention, a method of investigating a body by nuclear magnetic resonance comprises preferentially exciting nuclear magnetic resonance in a slice of the body; applying a magnetic field to said body with first and second gradients in first and second directions respectively in said slice, said first gradient being periodically reversed and said second gradient being applied only during periods when the first gradient has one of its two possible senses; detecting the free induction decay signals occurring during application of said magnetic field; assembling the resulting signals occurring during periods when said first and second gradients are both applied to form a first set of signals; assembling the resulting signals occurring during periods when only said first gradient is applied to form a second set of signals; processing the first set of signals by single Fourier transformation to derive data; and processing said second set of signals by double Fourier transformation to derive further data.
According to a third aspect of the present invention, a method of investigating a body by nuclear magnetic resonance comprises: preferentially exciting nuclear magnetic resonance in a slice of the body; applying a magnetic field to said body with first and second gradients in first and second directions respectively in said slice, said first gradient being periodically removed and then re-applied with its sense reversed, and said second gradient having one sense during each period when said first gradient is applied and being of the opposite sense at least during some of the periods when said first gradient is removed; and the magnitude of the second gradient being controlled so that the time integral of the second gradient at the beginning of each period when the first gradient has a given sense is substantially equal to the integral of the second gradient at the end of the preceding period when said first gradient had said given sense; detecting the free induction decay signals occurring during each period when said first gradient has said given sense, and processing the detected signals to retrieve data.
In order that the invention may be more fully understood reference will now be made to the accompanying drawings in which:
FIG. 1 shows alternative sampling techniques with square wave and sinusoidal modulation of a gradient,
FIG. 2 shows an experiment with cosinusoidal modulation and using non-linear sampling,
FIG. 3 shows an experiment with sinusoidal modulation and using non-linear sampling,
FIG. 4 shows a further experiment with cosinusoidal modulation and non-linear sampling,
FIG. 5 shows point spread sinc functions for cosinusoidal and sinusoidal experiments of equal period,
FIG. 6 is a diagram of a suitable circuit for driving a magnetic field gradient coil,
FIG. 7 shows a switching sequence that can be used with the circuit of FIG. 6,
FIG. 8 is a diagram of an alternative circuit to that shown in FIG. 6,
FIG. 9 shows a switching sequence applicable to the circuit of FIG. 8,
FIG. 10 shows another experiment enabling longer period examinations, and
FIG. 11 shows yet another experiment with cosinusoidal modulation.
Certain aspects of methods of investigating a body by nuclear magnetic resonance wherein, as in the methods of the present invention, the direction of the gradient of an applied magnetic field is repetitively reversed will first be discussed and explained.
Referring to FIG. 1, consider a sample or object being investigated which is placed in a static magnetic field B which extends in the Z-direction. The sample is subject to sequences of radio frequency(RF) fields and switched magnetic field gradients, as described in British patent specification No. 1,596,160. For reading out information from the object it is subjected to a read gradient Gy extending in the Y-direction which is repeatedly reversed with cycle time τ. If Gy is of constant amplitude as shown at FIG. 1(a) there is a constant evolution of the signal in spin phase and the Fourier transform (FT) of the signal is linear. The free induction decay (FID) signal and its echo as shown by the dotted line at FIG. 1(c) can then be regularly sampled corresponding to the constant evolution of the signal. However, if Gy varies sinusoidally, as for example illustrated in FIG. 1(d), then the FID signal appears as shown by the full line in FIG. 1(c) and the accumulated signal phase of the evolution signal must be the same when the FID has completely decayed to zero amplitude. However, in dephasing, it proceeds non-linearly as indicated in FIG. 1e. Take sample point A in FIG. 1b, where the gradient Gy has remained at its full value. For Gy varying as in FIG. 1d this point must be displaced to A' as in FIG. 1e. The signal dephasing and evolution slows down so that it reaches the same signal height later at point A' etc. This is quite general for any modulation Gy (t) and is not limited to sinusoidal modulation. If sampling is slowed down in unison with the gradient behaviour as indicated, then the accumulated phase of signal at each sample point will be the same as if the gradient were kept constant. Thus the sampling must be non-regular or non-linear but the FT is linear.
Let the accumulated phase of a particular volume element of spins at Δy in the steady gradient be
φn =γΔyGy τn (1)
if regularly sampled at time τn =nτo where γ is the magnetogyric ratio. Now if Gy (t) is some arbitrary function of time, we now require for each successive sampling step that: ##EQU1## where φ1 =γΔyGy τo, and now, of course τn is non-linearly stepped.
Thus for example if
Gy (t)=a cos wg t (3)
where wg is the angular frequency of the gradient modulation then sampling must follow a sine function, since ##EQU2##
Another way of looking at the problem is that the areas under the gradient curves for successive sample points must be in equal increments. The effect of this procedure is of course to lengthen the sampling process but the total length of the experiment can be retained at its original value by increasing the magnitude of Gy. For the sinusoidal or cosinusoidal cases this factor is π/2. An important consequence of non-regular sampling is that any originally linear evolution of the spin system now becomes non-linear following FT. Thus for example in echo planar imaging schemes described in British patent specification No. 1,596,160, the broadening gradient Gx would have to be modified because its role is to introduce linear broadening of the discrete profiles obtained by Gy modulation.
In general, then, if the experiment is performed off resonance with a broadening profile gradient Gx then this too must be modulated in unison with Gy but not reversed. That is to say, evolution in the same direction is required and that means that Gy =b cos wg t while Gx =c|cos ∩g t| where b and c are constants.
The non-linear sampling principle can be applied to all imaging schemes where linear gradients are required to be switched on and off or in any way periodically modulated. For example, non-linear signal sampling may be applied to advantage in (a) the projection reconstruction method where FID's are observed (following either a selective or non-selective 90°RF pulse) in a time-dependent read gradient which is in general easier to generate; (b) the Fourier imaging method described by Kumar, Welti and Ernst in which the total magnetisation following a short non-selective 90° pulse is allowed to evolve in three successively applied gradients Gx, Gy and Gz. In that version Gx, Gy and Gz are stepped square wave gradients, but provided accumulated total phase is stepped in equal amounts, that is to say (i) non-linear incrementing of the gradient period tx, ty is performed and (ii) non-regular sampling of the signal in the read gradient Gz is done, the problems of rapid gradient switching are resolved; (c) line scanning methods; (d) the multiple sensitive point method of Hinshaw which uses modulated gradients would also benefit by non-linear sampling. Attention is directed in this connection to U.K. Pat. No. 1,584,949.
Several embodiments of echo-planar imaging are possible. One embodiment, which involves cosinusoidal modulation of period 2τ, is described with reference to FIG. 2 and is a two-dimensional form which relies on an initial focussed selective irradiation RF pulse as shown at FIG. 2e to select a slice of spins. Following selective excitation in a magnetic field gradient Gz as shown at FIG. 2a, which is shown trapezoidal in this case because the rise, fall and reversal times of Gz times need not be particularly fast, read gradient Gy is switched on from zero, rampwise to its maximum value before going into a cosine modulation. The waveform of Gy isshown at FIG. 2b. The broadening field gradient Gx is likewise ramped up and then modulated according to a |cos wg t| form as shown at FIG. 2c. Alternative waveform shapes may be used to establish the starting conditions. For a given waveform four possibilities exist for the starting phases corresponding to either a positive or negative Gy gradient with a positive Gx gradient or a positive or negative Gy gradient with a negative Gx gradient.
The signals expected are sketched at FIG. 2f and the non-linear sampling which in this case varies as |sin wg t| is also shown at FIG. 2g. A different sampling pattern is required during ramp up to the cosine waveform if this particular starting condition is adopted.
If the experiment is performed on resonance a resonance offset magnetic field ΔB is not required since in that reference frame the spin system does not evolve. At resonance the received signal is detected in phase quadrature but this has a √2 S/N ratio advantage and is worth doing. The signal is recorded in phase quadrature, that is to say, in the rotating frame where both X and Y components of magnetisation are required using two analogue-to-digital converters (ADCs) recording simultaneously.
If it is desired to carry out an experiment off resonance then the offset magnetic field ΔB is also time modulated by a |cos wg t| modulation as shown at FIG. 2d. In order to carry out this modulation additional coils and driving circuits for the coils are provided. Such driving circuits supply current to the coils of a waveform corresponding to FIG. 2d.
Generation of a trapezoidal form selection gradient, (irregularities of which have the greatest effect as the selection gradient is switched off), may be obviated if the RF pulse shape is calculated using a non-linear transform which may be combined with a suitably modulated gradient to produce the correct selection profile. Alternatively the RF pulse may be generated linearly but applied in a non-linear manner, (the discreteness of the pulse varying in duration), allowing the gradient to be modulated as desired. Again, ΔB the resonance offset, must be varied in unison with the selection gradient, but working on resonance would remove this problem.
Any linear process like spin relaxation characterised by spin-lattice relaxation time T1, or spin-spin relaxation time T2, diffusion, and chemical shifts, which evolve while sampling non-linearly, will of course be non-linearised.
In the cosinusoidal modulation experiment described with reference to FIG. 2 the read gradient Gy is repetitively reversed to reform or refocus the signal repeatedly in order to impose a discreteness on the spin distribution. In the positive half cycles of the gradient Gy the Fourier transform that is obtained from the received signal from the object corresponds to the projection profile of the object along a line in one direction. However when the gradient is reversed the negative half cycles of the Gy gradient result in a projection profile along the same line in the opposite direction, that is to say it produces a mirror image of the profile.
In order to obtain a usable output, signals obtained during the negative half cycles of the gradient may be ignored. Alternatively, the signals obtained in the respective positive and negative Gy gradients can be subject to separate Fourier transforms and the frequency data of the signals obtained during the negative half cycles separately reordered. Both sets of frequency data can then be co-added together to form a single composite picture. Utilising both the positive and negative signals in this way enhances the signal-to-noise ratio of the final picture by a factor of √2. The experiment described above may be termed a "single shot" cosine experiment.
The data reordering described above is required whether the modulation is cosinusoidal or else has some other waveform that is symmetrical about the time origin.
According to one aspect of the present invention the experiment described above with reference to FIG. 2 is improved, as follows.
A first experiment is carried out as has been described with reference to FIG. 2 with an RF 90° selection pulse followed by a cosine modulated gradient +Gy as indicated by the solid line in FIG. 2b. If the period of the modulation of the Gy gradient is increased to 4τ an improvement in resolution is obtained. After a short delay the second experiment is carried out initiated by a second RF 90° selection pulse and this is followed by a gradient in the Y direction which is in antiphase to the original Gy gradient, that is to say, it is as shown by the chain dotted line in FIG. 2b. The purpose of the two separate experiments is to allow the nuclear signal responses to be edited to produce a first train of echoes occurring wholly in a positive Gy gradient and a second train of echoes occurring wholly in a negative Gy gradient. The edited FID signals are then Fourier transformed to produce non-distorted images from the positive and negative gradients. These images are then subsequently reordered and co-added to give a single composite image of improved signal-to-noise ratio.
The above pair of experiments may be termed a "double shot" experiment and may be carried out with selection pulses which are not 90° and may be performed either on resonance, in which case there is no field offset ΔB, or else off resonance.
As an alternative to the cosinusoidal modulation described with reference to FIG. 2, it is possible to utilise sinusoidal modulation of the magnetic field gradients with periodicity 2τ where all field modulations commence from zero and have sinusoidal waveforms throughout. Such an experiment is shown in FIG. 3. An initial RF pulse (FIG. 3e) is applied in the presence of a magnetic field gradient Gz of low value, which is then reversed (FIG. 3a). The broadening gradient Gx is then applied of form Gx =c |sin wg t| as shown at FIG. 3c and the read gradient Gy is applied as a sinusoidal modulation (FIG. 3b). A resonance offset field B (FIG. 3d) may also be applied if desired. The nuclear signal that is obtained is sketched at FIG. 3f. Since all gradients and field modulations start at zero magnitude, it is practically an easy business to switch on all modulations at t=0. The non-linear sampling as shown at FIG. 3g varies as predicted by a cosine function. A complementary experiment must also be performed with a Y-gradient modulation as indicated by the dashed line in FIG. 3b.
In the case of the sinusoidal modulation experiment described with reference to FIG. 3 the modulation of the Gy gradient is not symmetrical about the time origin and the Gy gradient changes direction during the formation of an echo signal. With such an arrangement sampling data obtained in the presence of a positive Gy gradient alone does not give an undistorted picture. Furthermore sampling data from positive half cycles and negative half cycles of the Gy gradient and reordering the latter also will not give a proper picture. Instead, to obtain a satisfactory picture data is sampled from positive half cycles of the Gy gradient alone in one experiment and then the experiment is repeated with the phase of the magnetic field gradient reversed as shown dotted at FIG. 3b. In the second experiment it is the now positive half cycles of the Gy gradient that are sampled corresponding to the times when the Gy gradient was negative in the first experiment. The two time signals are joined to form a single time echo sequence which is now wholly positive in evolution. Fourier transformation of this data yields a correct image. However generation of this image takes two experiments to produce rather than one. With moving objects where motion is asynchronous, as for example in cardiac motion and other involuntary bodily motions, delay between taking the positive phase and the negative phase sine-modulated signals could be unacceptable. It is therefore desirable for the two experiments to be carried out without delay between them. However so that some magnetisation remains in the selected slice for the second experiment it is convenient that the first selective pulse is not a 90° pulse but nutates through a smaller angle. A 45° nutation angle can be used and in this case the second selective phase may have a 90° nutation angle. In general nutation angles θ1, θ2 . . . . θn may be used for n looks at the spin magnetisation, where each pulse is optimised to give equal signals typically reduced from the signal obtained from a 90° nutation pulse by a factor 1/√n. Between the two experiments when there are no magnetic field gradients Gy or Gz in the Y or Z direction it may be advantageous to include a short intense magnetic field gradient in the X direction which further scrambles the decayed signal thus quenching any spin echo evolution which may form during the second experiment.
In order to avoid the necessity for carrying out two consecutive experiments when using cosinusoidal modulation of period 4τ it is possible, in accordance with a second aspect of the present invention, to carry out a modified form of experiment as illustrated in FIG. 5. A selective RF pulse shown at FIG. 4d is applied in the presence of a magnetic field gradient Gz to select a slice as before. After this selection a cosinusoidal magnetic field gradient Gy with a periodicity of 4τ is applied as shown at FIG. 4b in like manner to the similar waveform shown in FIG. 2b. However the broadening gradient applied in the X direction is provided in alternate half cycles only of the Gy gradient, as shown at FIG. 4c. In the intervening negative half cycles of gradient Gy no broadening gradient Gx is applied at all.
The signals obtained during the half cycles when both Gy and Gx are present are echo planar signals and these can be Fourier transformed in the manner described with reference to FIG. 2. In addition it will be found that nuclear signals will be generated in the intervening period when there is no Gx gradient. These signals can be ignored if desired but if it is so wished they can be utilised since they contain information relating to nuclear spin resonance in the selected slice. Each such signal is an FID signal which is sampled non-linearly and is of the form F(tm,tn) where tm are the non-equal time evolution periods of the spins in the presence of the -Gy gradient and tn refers to the linear accumulated spin phase in successive periods when the broadening gradient Gx is on. Double Fourier transformation of the signal F(tm,tn) yields a two-dimensional spin distribution of resonances in the selected slice. Thus in one experiment it is possible to obtain two complete images, namely an echo-planar image and a modified two-dimensional Fourier transform image.
The experiment described with reference to FIG. 4 may be done either on resonance as illustrated or else off resonance by the use of a magnetic field offset ΔB(t). However the off resonance case requires a precise setting of the amplitude of the static field offset so that there is an exact number of cycles of spin evolution in the periods when there is no magnetic field gradient Gx in the X direction.
The experiment described with reference to FIG. 4 makes the collection of echo-planar data take twice as long. This additional time need not be wasted since as described above additional Fourier transform data can be collected in that time. Nevertheless due to losses caused by spin-spin interactions which take place in a time T2 it may be impractical to double the experimental data gathering time. An alternative is to use a different negative waveform for the magnetic field gradient Gy to perform the spin echo refocussing. For example negative gradient spikes in place of the half sine curves would be equally efficacious provided that the total area under the spike is equal to that under the positive half sine. In the limit of a delta function spike the total experimental time decreases by a factor of 2.
An advantage with double shot cosine-modulated or sine-modulated experiments provided by the present invention is that the effect of truncation by gradient switcheing is minimised, since the gradient is in the same direction for twice as long as the single shot cosine-modulated experiments for equal modulation periods. The point spread sinc function, produced by the truncation, now falls to zero at the positions where successive stick spectra occur. This is illustrated by FIG. 5b which shows a spread function for a double shot sine experiment with the modulation period of Gy equal to 2τ. Thus, on broadening of the stick spectra, the resolution has been improved considerably along one axis of the picture. This resolution is to be compared with that of the single shot cosine experiment of FIG. 2. In this case the point spread sinc function is shown in FIG. 5a for the single shot cosine experiment of equal modulation period. The sinc zero crossings occur at twice the stick spectrum spacing, thus considerably degrading the point resolution between adjacent picture lines. However the same resolution can be obtained in a double shot cosine experiment running at half the frequency of the corresponding sine experiment.
An advantage of using sinusoidal or cosinusoidal modulation is that the coil arrangement by which a magnetic field gradient is applied to a sample can be made part of a tuned resonant circuit. This allows the current through the coil to be higher than the current supplied by a generator by a factor Q which is a quality factor of the tuned circuit and is give by Q=ω L/r, where ω is the angular resonant frequency, L is the inductance of the circuit and r is the series resistance. The use of a resonant circuit also allows switching of the current through the coil to take place at instants of zero current. A suitable circuit is shown in FIG. 6. It comprises a DC coupled audio frequency amplifier A the output of which feeds a capacitor C across which is connected an inductance L representing the gradient coil arrangement and a switch S. Switch S can be a mechanical, solid state or other electronic switch. Capacitor C is initially charged from amplifier A and when switch S is closed the capacitor discharges to generate sinusoidal current I through coil L of magnitude given by I=QId where Id is the drive current from amplifier A. By opening switch S when the current through coil L is zero, at which time the voltage across capacitor C is a maximum, and then closing switch S at a subsequent time separate half cycles of magnetic field can be produced of alternate directions separated from each other by the time during which switch S is open.
In certain circumstances a cosinusoidal current waveform is required commencing from a maximum value. While this is difficult to achieve in a practical arrangement a similar effect can be obtained by utilising the switching sequence shown in FIG. 7. The current through the gradient coil is shown at FIG. 7a. The cosinusoidal waveform proper commences at O' and is preceded by a quarter sine wave of current so that the complete current pulse is a half sine wave, which can be generated without difficulty. To obtain the required cosinusoidal-type effect the half-sine pulse is preceded by a prepulse of half-sine shape and of opposite polarity before launching into the sinusoidal modulation. The time integral of the prepulse is half that of the initial pulse proper. This ensures that at origin O' there is zero nett spin phase accumulation in Gy. The trigger periods during which switch S is closed is shown at FIG. 7b and the resulting voltage Vc across capacitor C is as in FIG. 7c. This prepulse facility for a cosine start can be used to advantage in previously described experiments embodied in FIG. 2 and FIG. 4, and in an experiment which will be described below in connection with FIG. 11.
As drawn all these Figures show an initial ramp-up to start off the cosine modulation for both Gy and Gx. While it is not difficult to achieve such a start for a small amplitude broadening gradient Gx such a ramp start for a larger amplitude gradient such as is applied in the Y direction can be difficult to achieve in practice. In such instances a prepulse is of value.
The circuit diagram illustrated in FIG. 6 can be expanded along the lines shown in FIG. 8. The gradient coil indicated as inductance L is now driven by two amplifiers A1 and A2 which each charge respective capacitors C1 and C2. The capacitors can be switched to discharge through the inductance L through one or other of two bi-directional switches S1 and S2. It is usually appropriate to make the two capacitors C1 and C2 of equal capacity and the two resonant circuits formed by C1 and L and C2 and L will have the same resonant frequency and the same value of Q. In order to generate the current waveform through inductance L shown at FIG. 9g drive currents D1 and D2 to the amplifiers A1 and A2 are as shown in FIGS. 9a and 9b respectively. The switching patterns for closing switches S1 and S2 are as shown at FIGS. 9c and 9d, where it will be noted that at no time are both switches closed simultaneously. The voltages Vc1 and Vc2 on the capacitors C1 and C2 are as shown in FIGS. 9e and 9f.
It may be noted that the gradient coil current waveform I although generated from a series of half sine waveforms, in fact has cosinusoidal symmetry and avoids the gradient prepulse discussed above. A disadvantage of using a "double bounce" cosine is that switching "glitches" occur at both zero crossing and at the waveform cusps. The cusp occurs at an echo maximum, but the gradient Gy is zero here, so spin evolution is relatively slow. Editing out cusp glitches in the signal may therefore be done by a simple interpolation procedure in which sampling points are moved to each side of the glitch and a mean value computed at the cusp itself.
An important advantage with either sinusoidal or cosinusoidal variation of gradients is that when used to investigate live patients switching transients in the patient are minimised and therefore risk reduced.
A limitation of the sampling time and hence signal-to-noise ratio in an echo-planar image is spin-spin relaxation, characterised by the parameter T2, which effects spin-echo recovery. A further practical limitation of pixel resolution and hence picture matrix size is the maximum amplitude of the magnetic field gradient Gy that may reasonably be generated. There is now described a method in which a single shot echo-planar experiment is broken down into a series of more manageable experiments with each individual experiment taking place in a time short enough to be relatively unaffected by spin-spin relaxation. The method described utilises non-linear sampling of the nuclear signals.
Since in general in echo-planar experiments Gy =lGx, where l (an integer) is the matrix size and further since the spatial resolution Δx is given by: Δx=2π/2lτγGx then for a given spatial resolution Δx and a longer time τ it is possible to reduce Gy. However for a given matrix size l and a given T2 such an increase in τ is limited. Breaking the experiment into shorter parts overcomes this problem.
The experimental sequence is described with reference to FIG. 10. An individual experiment commences with the application of a selective RF pulse in the presence of a magnetic field gradient Gz in the Z direction, followed by a negative Gz gradient. The waveform of the Gz gradients is shown at FIG. 10a and the RF pulse is shown at FIG. 10d. The RF pulse may be a 90° pulse or may have any other nutation angle O. In general it will be of the form R.sub.α (θ) where the subscript α refers to the RF carrier phase principle axis in the rotating reference frame. A cycle is terminated by the complementary selective RF pulse of waveform R-α (θ) in the presence of a magnetic field gradient Gz complementary to the initial Gz gradient. The pair of pulses R.sub.α (θ) and R-α (θ) constitutes a conjugate pulse pair having the effect of returning any transverse magnetisation back along the Z spin polarisation axis. The experiment may be repeated rapidly in the manner of a modified driven equilibrium Fourier transform (DEFT) experiment.
During a cycle a phase priming gradient Gx is initially applied in the X direction the time integral of which has a value An. This phase priming gradient is used to set up the spins in an orientation in which they would have existed at a particular time during the experiment had there been no effect due to T2, that is to say had the echo-planar gradient modulation been started earlier.
A magnetic field gradient Gy is then applied in the Y direction with a cyclic pattern of positive and negative gradients as set out in FIG. 10b. Each half cycle has a time period of 2τ. The nett effect of the application of the Gy gradient after the complete cycle illustrated has been carried out is zero. During the time that the read gradient Gy is applied the broadening gradient Gx in the X direction is also applied in synchronism therewith as shown in FIG. 10c. The Gx gradient also consists of a sequence of half cycles, the half cycles being positive in the first half of the experiment and negative in the second half or vice versa. After the complete cycle of the application of the Gy gradients is completed a negative Gx gradient of time integral An is then applied. This refocusses the spins and then the experiment is terminated by the second pulse of the conjugate pair of RF pulses which is applied in the presence of the complementary Gz gradient field in order to return the spins predominantly along the Z axis.
After a suitable wait period the experiment described above may be repeated so as to give a continuously running method of collecting data. The value of An in a particular experiment is given by
where ΔA=4Ao as illustrated, but could include m sub-cycles, in which case ΔA=4mAo, where m is an integer 1, 2, 3 etc. and Ao is the time integral of any one half period 0, 1, 2, etc. and is stepped for each successive experiment so that a different part of the time domain is investigated in turn.
The above sequence of experiments has been described as being carried out without a field offset to the static magnetic field. If it is desired to carry out a sequence of experiments off resonance then a field offset is required which is varied in synchronism with the Gx gradient field. The value of the field offset prepulse is different for each successive experiment and in general has a value, Bn
where Bn =nΔB
and ΔB=4mBo ; and
Bo is the signal evolution in a half period due to the field offset term and as illustrated m=1.
The waveforms of the time variation of the read gradient Gy and broadening gradient Gx are shown in FIG. 10 as having triangular shapes. However the arrangement described with reference to FIG. 10 is in no way limited to triangular modulation but half sine or trapezoidal modulation can be used or in fact any other repetitive waveform that it is convenient to generate. Furthermore any suitable arrangement for gradient switching can be used.
An improvement of echo-planar imaging in accordance with the third aspect of the invention will now be described with reference to FIG. 11. This particular method is a single shot experiment with cosinusoidal modulation and it has a spatial resolution equal to that of the double shot experiment with sinusoidal modulation. Selection of a slice or plane is performed as described previously by the application of a selective RF pulse shown in FIG. 11e in the presence of a magnetic field gradient Gz in the Z direction as shown in FIG. 11a. Gradient Gz has trapezoidal modulation and is followed by a reversed gradient field Gz of half the time integral to give focussed magnetisation. The read gradient Gy is initially ramped up and is then cosine modulated with a period for each half cycle of 4τ. However between each half cycle there is a small time gap Δτ during which no gradient Gy is applied at all. The waveform of the gradient Gy is shown in FIG. 11b. The PG,24 broadening gradient Gx in the X direction is applied in synchronism with the Gy gradient and is in the same direction (positive in FIG. 11c) for both the positive and negative excursions of the Gy gradient. However in the time gaps Δτ when the Gy gradient is zero the broadening gradient Gx is applied in the reverse direction as a half-sine pulse or spike with a magnitude such that the time integral after the first pulse is equal to the time integral of the immediately preceding magnetic field gradient Gx in the positive X-direction (see hatched areas in FIG. 11c). The effect of this equalisation of time integrals is to exactly cancel the spin phase evolution in the X direction at the end of the spike. In the following half cycle of modulation of time 4τ during which the read gradient field Gy is in the negative direction the spin phase in the X direction will evolve afresh. At the end of this half cycle there will be another reversed spike of magnetisation in the - X direction of time integral equal to the previous spike. At the end of this second spike therefore only half the phase evolution is cancelled so that the spin phase in the X direction is prepared in a state equivalent to that which would have occurred if the negative evolution of spins in the -Y direction had not occurred. That is to say sampling just the even echoes which occur during the application of the positive gradient Gy results in a correctly ordered single shot signal which when Fourier transformed yields an undistorted image with equivalent sine resolution. The FID signal and echoes that are obtained are illustrated in FIG. 11f. These signals are sampled non-linearly as previously described.
As a modification to the scheme described with reference to FIG. 11 the spikes of broadening gradient Gx which are applied in the -X direction can have a time integral double the time integral of the initial cosine modulation of Gx but these spikes are then applied only during alternate time gaps in the modulation. This is shown in FIG. 11d. It is desirable to keep the width of the time gap Δτ as short as possible in order to keep the total experimental time short.
It will be appreciated that whilst the methods described above have been described in the content of obtaining an image or other indication of nuclear magnetic resonance spin density in a sample, the methods according to the present invention are equally applicable to other methods of investigating a body e.g. methods of obtaining an indication of the spatial distribution of other chosen NMR quantities, e.g. NMR spin relaxation time constants or chemical shift.
Similarly whilst the methods according to the invention have been described in the context of investigating a slice of a body, the methods may equally be used in investigating a volume of a body, i.e. a plurality of adjacent slices simultaneously.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4318043 *||Feb 12, 1980||Mar 2, 1982||The Regents Of The University Of California||Method and apparatus for rapid NMR imaging of nuclear densities within an object|
|US4355282 *||Jul 29, 1980||Oct 19, 1982||Picker International Limited||Nuclear magnetic resonance systems|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US4588948 *||Oct 4, 1983||May 13, 1986||Peter Mansfield||Nuclear magnetic resonance methods|
|US4618827 *||Sep 10, 1984||Oct 21, 1986||General Electric Company||Method for high-spatial-resolution spectroscopic NMR imaging of chemically-shifted nuclei|
|US4628264 *||Mar 14, 1984||Dec 9, 1986||Advanced Nmr Systems, Inc.||NMR gradient field modulation|
|US4644282 *||May 23, 1984||Feb 17, 1987||Siemens Aktiengesellschaft||Apparatus for the formation of images of an examination subject with nuclear magnetic resonance|
|US4647857 *||Oct 22, 1984||Mar 3, 1987||Stanford University||Flow measurement using nuclear magnetic resonance|
|US4651096 *||Feb 19, 1985||Mar 17, 1987||Resonex||Imaging method for nuclear magnetic resonance utilizing balanced gradient fields|
|US4678996 *||May 7, 1985||Jul 7, 1987||Picker International, Inc.||Magnetic resonance imaging method|
|US4689560 *||Nov 15, 1985||Aug 25, 1987||Picker International, Inc.||Low R.F. dosage magnetic resonance imaging of high velocity flows|
|US4689562 *||Oct 11, 1983||Aug 25, 1987||Elscint Ltd.||NMR Imaging method and system|
|US4689568 *||Jul 1, 1985||Aug 25, 1987||Hitachi, Ltd.||NMR chemical shift imaging method|
|US4707660 *||Aug 22, 1985||Nov 17, 1987||U.S. Philips Corporation||Fast method and device for determining an NMR distribution in a region of a body|
|US4709211 *||Jul 11, 1986||Nov 24, 1987||Kabushiki Kaisha Toshiba||Nuclear magnetic resonance system|
|US4716369 *||Jun 20, 1985||Dec 29, 1987||Hitachi, Ltd.||High speed imaging method with three-dimensional NMR|
|US4733188 *||Apr 10, 1986||Mar 22, 1988||Hitachi, Ltd.||NMR imaging method|
|US4740748 *||Dec 3, 1986||Apr 26, 1988||Advanced Nmr Systems, Inc.||Method of high-speed magnetic resonance imaging|
|US4760339 *||Dec 22, 1986||Jul 26, 1988||Hitachi, Ltd.||NMR imaging method|
|US4812762 *||Dec 22, 1987||Mar 14, 1989||U.S. Philips Corp.||Fast method and device for determining an NMR distribution in a region of a body|
|US4837513 *||Dec 5, 1988||Jun 6, 1989||The Regents Of The University Of California||MRI using asymmetric RF nutation pulses and asymmetric synthesis of complex conjugated SE data to reduce TE and T2 decay NMR spin echo responses|
|US4845430 *||Jun 29, 1987||Jul 4, 1989||Kabushiki Kaisha Toshiba||Magnetic resonance imaging system|
|US4891594 *||Aug 19, 1988||Jan 2, 1990||Resonex, Inc.||Method of optimizing flip angles in MRI for unequal delay times|
|US4893081 *||Mar 30, 1989||Jan 9, 1990||Elscint Ltd.||Driven equilibrium in magnetic resonance imaging|
|US4940941 *||Apr 3, 1989||Jul 10, 1990||Advanced Nmr Systems, Inc.||Method of high-speed magnetic resonance imaging|
|US5079503 *||Mar 26, 1990||Jan 7, 1992||Siemens Aktiengesellschaft||Magnetic resonance imaging apparatus and method for operating same|
|US5119027 *||Aug 14, 1989||Jun 2, 1992||Ehrhardt James C||Nuclear magnetic resonance imaging method|
|US5162736 *||Jan 18, 1991||Nov 10, 1992||National Research Development Corporation||NMR imaging|
|US5233301 *||Aug 16, 1990||Aug 3, 1993||Board Of Trustees Of The Leland Stanford Junior University||High resolution/reduced slice width magnetic resonance imaging and spectroscopy by signal combination and use of minimum phase excitation pulses|
|US5252923 *||Jun 29, 1990||Oct 12, 1993||British Technology Group Limited||Nuclear magnetic resonance imaging methods|
|US5285161 *||Jun 1, 1993||Feb 8, 1994||Advanced Nmr Systems, Inc.||Circuitry for driving field-generating coil of magnetic resonance imaging system|
|US5287854 *||Apr 12, 1990||Feb 22, 1994||Nycomed Innovation Ab||Electron spin resonance enhanced MRI using an echo planar imaging technique|
|US5347217 *||Jul 25, 1991||Sep 13, 1994||British Technology Group Limited||Magnetic resonance spectroscopy and imaging|
|US5623206 *||Jul 29, 1994||Apr 22, 1997||British Technology Group Limited||Nuclear magnetic resonance (NMR) imaging|
|US5652516 *||Jan 22, 1996||Jul 29, 1997||The Board Of Trustees Of The Leland Stanford Junior University||Spectroscopic magnetic resonance imaging using spiral trajectories|
|US6043652 *||Apr 17, 1997||Mar 28, 2000||Picker International, Inc.||Alternative reconstruction method for non-equidistant k-space data|
|US6215309||Dec 1, 1994||Apr 10, 2001||Aurora Imaging Technology, Inc.||Circuitry for driving field-generating coil of magnetic resonance imaging system|
|US6219571||Mar 29, 1999||Apr 17, 2001||Board Of Trustees Of The Leland Stanford Junior University||Magnetic resonance imaging using driven equilibrium fourier transform|
|US6831459 *||Jul 11, 2003||Dec 14, 2004||The Research Foundation Of State University Of New York||Method of using G-matrix Fourier transformation nuclear magnetic resonance (GFT NMR) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US7365539||Oct 26, 2004||Apr 29, 2008||The Research Foundation Of State University Of New York||Method of using G-matrix fourier transformation nuclear magnetic resonance (GFT NMR) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US7719274 *||Jan 19, 2007||May 18, 2010||Korea Advanced Institute Of Science & Technology||Method of obtaining a magnetic resonance image in which the streak artifacts are corrected using non-linear phase correction|
|US7859263||Jan 7, 2009||Dec 28, 2010||Ge Medical Systems Global Technology Company, Llc||Magnetic resonance imaging apparatus and program, and RF pulse transmission method|
|US7880468||Mar 27, 2008||Feb 1, 2011||The Research Foundation Of State University Of New York||Method of using G-matrix fourier transformation nuclear magnetic resonance (GFT NMR) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US7920972||Oct 18, 2005||Apr 5, 2011||The Research Foundation Of State University Of New York||G-matrix Fourier transformation (GFT) nuclear magnetic resonance (NMR) experiments for resonance assignment and structure determination of organic molecules|
|US8487614 *||Dec 10, 2009||Jul 16, 2013||Kabushiki Kaisha Toshiba||Magnetic resonance imaging apparatus/method counter-actively suppressing remnant eddy current magnetic fields generated from gradients applied before controlling contrast pre-pulses and MRI image data acquisition|
|US9594144||Apr 23, 2014||Mar 14, 2017||General Electric Company||Low-noise magnetic resonance imaging using low harmonic pulse sequences|
|US20040061497 *||Jul 11, 2003||Apr 1, 2004||Szyperski Thomas A.||Method of using G-matrix Fourier transformation nuclear magnetic resonance (GFT NMR) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US20050114038 *||Oct 26, 2004||May 26, 2005||Szyperski Thomas A.||Method of using G-matrix fourier transformation nuclear magnetic resonance (GFT NMR) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US20050244929 *||Jul 1, 2005||Nov 3, 2005||Genentech, Inc.||Expression of functional antibody fragments|
|US20090009166 *||Mar 27, 2008||Jan 8, 2009||The Research Foundation Of State University Of New York||Method of using g-matrix fourier transformation nuclear magnetic resonance (gft nmr) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|US20090179641 *||Jan 7, 2009||Jul 16, 2009||Tsuyoshi Matsuda||Magnetic resonance imaging apparatus and program, and rf pulse transmission method|
|US20100085049 *||Jan 19, 2007||Apr 8, 2010||Korea Advanced Institute Of Science & Technology||Method of obtaining a magnetic resonance image in which the streak artifacts are corrected using non-linear phase correction|
|US20100148774 *||Dec 10, 2009||Jun 17, 2010||Mitsukazu Kamata||Magnetic resonance imaging apparatus and magnetic resonance imaging method|
|USRE32701 *||Apr 8, 1987||Jun 21, 1988||Wisconsin Alumni Research Foundation||NMR scanner with motion zeugmatography|
|WO1985004258A1 *||Mar 7, 1985||Sep 26, 1985||Advanced Nmr Systems, Inc.||Nmr gradient field modulation|
|WO2004007016A2 *||Jul 10, 2003||Jan 22, 2004||The Research Foundation Of State University Of New York||A method of using g-matrix fourier transformation nuclear magnetic resonance (gft nmr) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|WO2004007016A3 *||Jul 10, 2003||Oct 14, 2004||Univ New York State Res Found||A method of using g-matrix fourier transformation nuclear magnetic resonance (gft nmr) spectroscopy for rapid chemical shift assignment and secondary structure determination of proteins|
|WO2015164169A1 *||Apr 16, 2015||Oct 29, 2015||General Electric Company||Low-noise magnetic resonance imaging using low harmonic pulse sequences|
|U.S. Classification||324/309, 324/307|
|International Classification||A61B10/00, G01R33/561, A61B5/055, G01R33/48|
|Cooperative Classification||G01R33/5616, G01R33/5615|
|Aug 27, 1985||CC||Certificate of correction|
|Sep 1, 1988||FPAY||Fee payment|
Year of fee payment: 4
|Sep 21, 1992||FPAY||Fee payment|
Year of fee payment: 8
|Aug 29, 1996||FPAY||Fee payment|
Year of fee payment: 12